“Characterization of a functional NTPDase in the ... · Submandibular glands (SMG) belong to the...
Transcript of “Characterization of a functional NTPDase in the ... · Submandibular glands (SMG) belong to the...
“Characterization of a functional NTPDase in the endoplasmic reticulum of
rat submandibular salivary gland”
Mariano Anibal OSTUNI.a,b,*, Patricia EGIDO b, Gabriel PERANZIa, Guillermo Luis
ALONSOb, Jean-Jacques LACAPEREa., and Débora Alejandra GONZALEZb
a INSERM U773, Centre de Recherche Biomédicale Bichat-Beaujon (CRB3), Université
Paris 7 Denis Diderot, site Bichat . 16, rue Henri Huchard, 75870 Paris Cedex, France. b Cátedra de Biofísica, Facultad de Odontología, Universidad de Buenos Aires. M.T. de
Alvear 2142 C1122AAH, Buenos Aires, Argentina.
*Corresponding author:
Mariano A. Ostuni. INSERM U773, Centre de Recherche Biomédicale Bichat-Beaujon
(CRB3). 16, rue Henri Huchard, 75870 Paris Cedex, France. Tel (+33) 1 57 27 74 80. Fax
(+33) 1 57 27 74 80. E-mail address: [email protected]
Running Title: NTPDase activity in submandibular gland
1
Summary.
Nucleotidase activity and Ca-uptake were characterized in endoplasmic reticulum (ER)
enriched rat submandibular gland (SMG) microsomal preparations. (i) Ca-uptake presented
characteristics of an ER Ca-ATPase. (ii) Nucleotidase activity was equally stimulated by
calcium, magnesium and manganese, but with different Km values. (iii) Specific inhibitors of
P-type Ca-ATPases were ineffective on nucleotidase activity, demonstrating that this activity
was not related to calcium uptake and did not correspond to classical Ca2+ pumps. (iv) ATP
and UTP were more efficient substrates, whereas ADP and UDP were hydrolyzed with
significantly slower rate. (v) Nucleotidase activity was sensitive to mild detergent
solubilization and insensitive to ionophore addition. (vi) Nucleotidase activity was strongly
inhibited by suramin, a nucleoside triphosphate diphosphohydrolase (NTPDase) inhibitor.
(vii) Nucleotidase activity exponentially diminished as function of time. All these
observations are consistent with a NTPDase identity. The presence of a NTPDase was
demonstrated by immunohistochemistry in rat SMG. Immunoreactivity was stronger in ductal
cells than in mucus and serous acini. Although this enzyme was observed in the plasma
membrane, colocalization with the ER marker calnexin revealed a specific subcellular
localization in this organelle of all three types of cell. The putative function of this NTPDase
activity in salivary glands is discussed.
Key words: Apyrase, CD39, Ca-uptake, ATPase activity, submandibular gland.
2
Introduction
Submandibular glands (SMG) belong to the major salivary gland group, with important
functions involved in the oral health and the digestion process. They are widely used as a
model to study the secretory mechanisms in the epithelial cells.
Regulation of salivary secretion is complex and secretory agents usually act by cAMP and
Ca2+ pathways, which could regulate the same process with opposite or synergic results.
The Ca2+ pathway implies the fine-tuning regulation of cytosolic calcium levels. Ca2+ influx
from the extracellular liquid and Ca2+ release from intracellular calcium stores are strongly
related with the cation extrusion and the stores refilling. Several pumps and calcium channels
have been identified in submandibular glands. The group of S. Muallem has identified in the
rat submandibular gland different calcium ATPases as the plasma membrane Ca-ATPase
(PMCA) and the sarco-endoplasmic reticulum Ca-ATPase (SERCA) (Lee et al 1997a), and
calcium channels like the IP3-sensitive and the ryanodyne-sensitive (Lee et al 1997b).
The functional characterization of the SERCA was usually addressed measuring the
enzymatic activity and the ionic transport in submandibular endoplasmic reticulum
microsomes (Alonso et al 1971; Hurley and Martínez, 1985, 1986; Hurley and Ryan 1988),
but a clear correlation between Ca2+ uptake and ATPase activity was never well established.
At present the question remains what is the actual identity of the enzymes that hydrolyze ATP
in those microsomes.
Another class of ATP hydrolyzing enzymes is the family of ecto-nucleoside triphosphate
diphosphohydrolases (NTDPases) that hydrolyze ATP and other nucleotides with largely
higher hydrolysis rate than the Ca-ATPase pumps, but without coupled cation transport (for
detailed review see Plesner 1995; Robson et al 2006). At present, eight members of this
family were idientified, which are addressed to the plasma membrane and/or to intracellular
membranes (Robson et al 2006).
NTPDases are ubiquitous in eukaryotic cells but their functions are not fully understood.
Cumulate evidences support the involvement of these enzymes in the regulation of nucleotide
concentration and adenosine recycling process. Furthermore their spatial and temporal
expression in several cell types could be at the origin of different patho-physiological
processes (Reviewed in Robson et al 2006).
Several articles characterizing an NTPDase like activity in the rat parotid gland were
published (Dodw et al 1989, 1993, 1996, 1999; Teo et al 1993; Murphy et al 1994), but there
are only a cup of papers talking about this kind of activity in rat submandibular gland
(Valenzuela et al 1989, Murphy et al 1994)
3
In this work we characterize an ATPase activity from endoplasmic reticulum microsomes
from rat submandibular gland, which has the characteristics of an ecto-nucleotidase activity
and that it is not related to calcium uptake. We also identify the cellular and subcellular
localization of a NTPDase immunoreactivity in this gland.
4
Methods
Animals
Male Wistar rats weighing 250 to 280 gr. were fed with standard chow diet and water ad
libitum, and kept with 12 h light / 12 h dark period. They were treated in accordance with the
European Convention for the Protection of Vertebrate Animals used for Experimental and
other Scientific Purposes (CETS 123).
Microsomes preparation
Microsomes preparation protocol was modified from previously described (Hurley and
Martinez 1986). Briefly, rats were allowed free access to standard chow pellets and water
except for the night before sacrifice when only food was withheld. For each tissue preparation
2 to 4 rats were deeply anaesthetized by exposure to air saturated with ether and
exsanguinated via the abdominal aorta. The submandibular glands were removed, separated
from the adjoining sublingual glands, weighed, minced on ice and homogenized in cold buffer
(10 mM HEPES, pH 7.0, 0.3 M sucrose and 1 mM of the protease inhibitor PMSF) using a
teflon and glass homogenizer. The homogenate was centrifuged for 2 minutes at 1000 x g to
remove nuclei and cell debris. Supernatant was centrifuged for 20 minutes at 10,000 x g, the
pellet containing mitochondrial fraction was discarded and resulting supernatant was
centrifuged for 30 minutes at 27,000 x g. The final pellet was re-suspended in the same buffer
and proteins were quantified using the dye-binding BioRad assay.
Electron Microscopy
Microsome preparation was fixed in 2.5% glutaraldehyde during 60 min at room temperature
and postfixed in 1% osmium tetroxide for 60 min. It was dehydrated and embedded in LX-
112 resin and sections were contrasted with 2% uranyl acetate and observed in a JEOL 1200
EX electron microscope operated at 80 kV accelerated voltage.
Electrophoresis and Western blotting
SDS-polyacrylamide gel electrophoresis was performed according to Laemmli (Laemmli
1970). Proteins (15 µg per lane) were separated on 10 % polyacrylamide gel under non
reducing conditions. They were transferred to 0.45 µm pore nitrocellulose membranes
(Sartorius AG, Göttingen, Germany) and probed with monoclonal anti-hNTPDase1 antibody
(BU61 clone, Ancell, Bayport, MN, USA) (1:750 dilution); or with a rabbit anti-calnexin
monoclonal antibody (Becton Dickinson, Le Pont de Claix, France). (1:4000 dilution). Bands
were visualized using peroxidase-conjugated anti-mouse or anti-rabbit IgG respectively
(Sigma, Saint-Quentin Fallavier, France) and the Pierce ECL western blotting substrate
(Thermo Scientific, Rockford, IL, USA).
5
Deglycosylation
Microsome preparations were subjected to deglycosylation using PNGase F (New England
Biolab, Ipswich, MA, USA) following manufacturer instructions.
ATPase activity
The microsomal ATPase activity was measured at pH 7.2 and 37°C following Pi liberation.
The media contained 50 mM Mops-Tris, pH 7.2, 100 mM KCl, 3 mM ATP and different
amounts of CaCl2 or MgCl2, and EGTA to obtain the desired cation concentration. Reactions
were started by addition of 0.1 mg vesicular protein per ml. After 2 min incubations the
reactions were stopped by addition of cold trichloroacetic acid at 5% final concentration. Pi
production was determined by a colorimetric method (Baginski et al 1967). Free cation
concentrations were calculated as previously described (Fabiato and Fabiato 1979).
[45Ca]Ca-uptake
Ca-uptake was measured in 1 ml of medium of the following composition: 100 mM KCl, 50
mM MOPS/Tris, pH 7.2, 3 mM ATP, 3 mM MgCl2, 0.1 mM CaCl2, also containing 1500 ±
500 cpm/nmol of [45Ca] (New England Nuclear, Boston, MA, USA), and 0.1 mg of
membrane protein. The reaction mixture lacking membrane proteins was pre-incubated for 5
min at 37 °C. Ca-uptake was started by addition of membrane proteins and aliquots of the
reaction mixture were passed through Millipore filters (0.45 µm pore diameter) at the given
time periods. The filters were immediately washed with 5 ml of 3 mM LaCl3 in order to
diminish the unspecific binding. Radioactivity was determined by liquid scintillation counting
in a toluene-based scintillation cocktail.
Immunohistochemistry
Submandibular glands from 4 rats were removed and cleaned from connective tissue. They
were immediately fixed in 4% paraformaldehyde during 4 hours at room temperature and
embedded in paraffin. Paraffin sections were re-hydrated and endogenous peroxidase activity
was inactivated followed by the peroxidase-antiperoxidase (PAP) technique. Sections were
incubated with a mouse anti-human NTPDase1 (hNTPDase1) monoclonal antibody (BU61
clone, 1:100 dilution, Ancell, Bayport, MN, USA) overnight at 4°C. Peroxidase complexed
secondary antibody was purchased from Dakopatts Ltd. (Trappes, France). Sections were
counterstained with Mayer’s hematoxilin (E.M.S.) (Euromedex, Souffelweyersheim, France).
Colocalization Studies
Paraffin sections for confocal microscopy were re-hydrated and incubated with a rabbit anti-
calnexin monoclonal antibody (1:100) for 1h (Becton Dickinson, Le Pont de Claix, France).
After washing, sections were incubated with a secondary anti-rabbit immunoglobulins
6
antibody coupled with ALEXA Fluor 488 (1:200) (Molecular Probes, Oregon, USA) and then
incubated overnight at 4°C with the anti-hNTPDase1 antibody (1:100). Following second
washing, secondary anti-mouse immunoglobulins antibody coupled with ALEXA Fluor 546
(Molecular Probes, Oregon, USA) was deposited for 60 min (1:150 dilution). After several
washes, slides were covered with a hydrophilic and antifading mounting media (Fluoromount
G, E.M.S.). Final observations were made with a Zeiss Axiovert 200M confocal microscope
with a 63x oil immersion objective. LSM 510 image software (Heidelberg, Germany) was
used to select observation fields (identical intensities profile of each fluorochrome), to collect
the results (histograms, fluorograms, colocalizations of both dyes), and to derive data
recording and analysis. Step size for the sectioning was 0.4 µm for the deconvolved images
with a pinhole value fixed at 1 µm. Similar incubations without the primary antibody were
also performed as negative controls.
Reagents and solutions
C12E8 was purchased from Calbiochem (La Jolla, CA, USA). Unless otherwise noted, all
other reagents were purchased from Sigma Chemical Co. (St. Louis, MO, USA).
Statistics
Data are presented as the mean ± standard error of mean (SEM). Experimental data were
studied using one-way ANOVA. Scheffe homogeneity test was performed in order to
determinate differences between means of each experimental group and control counterpart.
Differences were considered significant for p< 0.05.
7
Results Electron Microscopy and western blot.
Figure 1a shows typical image of section from microsomal preparation included in resin.
Vesicles surrounded by large dots corresponding to ribosomes are clearly observed and
characteristics of rough ER microsomes. Dotted surfaces are also observed and correspond to
surfaces of vesicles that have not been sectioned. It has to be mentioned that neither
mitochondria nor mitochondrial fragments were observed in these microsomal preparations.
The ER origin of these microsomes was confirmed by western blot with the presence of a
characteristic band of 88 kDa that corresponds to the ER specific protein calnexin, (Fig. 1b).
[45Ca]Ca-Uptake.
Figures 1c and 1d show time dependent Ca-uptake from SMG microsomes. Panel c
demonstrates ATP dependence of this uptake, resulting in an increase of calcium inside the
vesicles, observed only in the presence of ATP. A fast release of Ca2+ was triggered by
addition of Ca2+ ionophore A23187..
The effect of two SERCA inhibitors on Ca-uptake are shown in the Fig 1d. Thapsigargin and
sodium orthovanadate diminished the Ca2+ uptake from SMG microsomes by about 90 %.
The mitochondrial Ca-uniporter inhibitor ruthenium red and the PMCA inhibitor compound
48/80 had no effect on SMG microsomes Ca2+ uptake (data not shown). These are the first
results about the ATPase inhibitors action on the rat submandibular Ca2+ uptake. They
confirm the microsomal origin of the ER membranes while minimizing the possibility of
mitochondrial or plasma membrane contamination.
ATPase Activity
To functionally characterize the ATPase activity in our micromal preparation, we tested: i)
the dependence on Ca2+ and/or Mg2+, (ii) the sensitivity to specific inhibitors of P-type, F-
type, and V-type ATPases, (iii) the substrate preference pattern, (iv) the sensitivity to the
action of detergents and (v) the dependence on incubation time.
Figure 2a shows ATPase activity of SMG microsomes as function of free Mg2+ with no added
calcium or free Ca2+ without added magnesium. Maximal activities were similar for both
conditions (around 1 µmol Pi produced/mg protein/min) but the Km values calculated from
plots were different with values of 28 ± 5 µM and 320 ± 51 µM for Ca2+ and Mg2+,
respectively. ATPase activity was also stimulated by Mn2+ (open squares). Maximal activity
was reached in the presence of mM concentration of Ca2+ or Mg2+ irrespective of other cation
concentration (data not shown).
8
Table 1 shows the effect of different ATPase inhibitors. No significant inhibition of ATPase
activity was obtained by the SERCA inhibitor thapsigargin, the P-type ATPases inhibitor
sodium orthovanadate, the PMCA inhibitor compound 48/80, or the Na/K ATPase inhibitor
ouabain. However, sodium azide, an unspecific NTPDase inhibitor (Knowles and Nagy 1999;
Schetinger et al 2001), slightly diminishes ATPase activity (67.4 % of control), while
suramin, a NTPDase inhibitor (Iqbal et al 2005), produces a strong inhibition of this activity.
Figure 2b showed the concentration-dependent inhibition profile from which IC50 could be
calculated giving values of 55 ± 10 µM and 585 ± 160 µM for suramin and sodium azide,
respectively. The simultaneous addition of suramin and sodium orthovanadate at saturating
concentrations produced an additional 5 % inhibition, whereas addition of calcium ionophore
A23187 did not change ATPase activity.
Hydrolysis of different tri- and di- nucleotides are presented in figure 2c. The SMG
microsomes nucleotidase activity is principally a tri-nucleotidase activity. ATP and UTP are
largely hydrolyzed compared to ADP or UDP. ATP/ADP and UTP/UDP hydrolysis ratio are
4.39 and 4.93, respectively. Figure 2d shows that ATPase activity decayed exponentially.
Differential activity values were obtained from the derivative of the Pi production as function
of time (not shown)
These results strongly suggest that this ATPase activity resulted from an NTPDase. Moreover,
addition of detergent C12E8 significantly inhibited the ATPase activity of SMG microsomes
(28 % of control, table 1). If this ATPase activity was the product of a SERCA type enzyme,
mild detergent solubilization would augment activity by abolishing inhibition by
intravesicular Ca2+ concentration. Furthermore, it has been previously described that detergent
had a strong inhibitory effect on NTPDase activity (Wang and Guidotti 1998).
Cation and substrate specificity, inhibitors effect, and detergent sensitivity, indicate the
presence of one or several NTPDases, probably belonging to the NTPDase1 related enzymes
(NTPDases 1, 2, 3 and 8), which are usually located in the plasma membrane. However
different articles reported NTPDase1 activity in subcellular compartments in eukaryotic cells
(Zhong et al 2001; D’Alessio et al 2003) opening the question about the possible intracellular
function of these enzymes.
Immunohistochemistry:
Based on the highly conserved amino acid sequence of different NTPDases among species,
we tested the specificity of a commercially available anti human NTPDase1 (hNTPDase1)
monoclonal antibody.
9
Western blot performed from non reducing conditions PAGE revealed that this antibody was
able to recognize a band of 78 kDa both in human platelets lysate as well as in rat whole
homogenate and microsomal fraction (Fig. 3a, lanes 1-3). It’s noteworthy that 54 kDa bands
are proportionally most important in SMG microsomal fraction compared to SMG whole
homogenate, indicating that microsomal fraction presented a significant amount of non
glycosylated enzyme (Lemmens et al 2000). After PNGase F digestion, microsomal fraction
showed a reduced 78 kDa band and the presence of a new band of about 28 kDa (Fig 3 lane
4).
Figure 3b-e shows rat SMG immunoreactivity against anti-hNTPDase1 monoclonal antibody.
As expected, immunoreactivity was strongly revealed in blood vessels, which are well known
to express high levels of NTPDases. Cells of epithelial ducts (Figs. 3b,c) and mucous acini
(Figs. 3b,d) are homogeneously labelled, whereas only 20% of serous acini were anti
hNTPDase1 positive (Figs. 3b,e). Figures 3c-e reveal that immunoreactivity was located both
in the plasma membrane and in the cytosol of rat SMG cells with stronger reactivity in the
plasma membrane of luminal pole (Figs. 1c-e, arrows heads). NTPDase immunoreactivity
was principally located in the luminal pool of duct cells, whereas labeling in serous and
mucous acini is distributed through the cells. This difference might be accounted by the
nuclear localization of duct cells near the luminal membrane, while acinar cells nuclei are
distributed throughout the cytoplasm.
Distribution of NTPDase immunoreactivity presented herein was different to one described in
the mouse SMG where mucous acini were weakly stained while duct cells were not reactive
to anti CD39 polyclonal antibody at all (Kittel et al 2004).
Immunofluorescence and colocalization
Present observations of NTPDase1-like immunoreactivity in the cytosol (Fig. 3c-e) suggest
localization in the endoplasmic reticulum. This hypothesis was tested by immunofluorescence
colocalization experiments using two antibodies reacting to NTPDase1 and to calnexin
(specific endoplasmic reticulum protein (Bergeron et al. 1994)). In figure 4 anti calnexin
immunoreactivity (a,d,g), anti NTPDase1 labeling (b,e,h) and colocalization images (c,f,i)
from duct cells (a,b,c), mucous acini (d,e,f) and serous acini (g,h,i) are respectively shown.
The three cell types show positive anti-NTPDase immunofluorescence (red) and a strong
reaction to anti-calnexin antibodies (green). Anti-calnexin immunoreactivity presents the
typical perinuclear pattern of ER localization.
Colocalization analysis was made using the same fluorescence intensity for both signals and
pixel-to-pixel superposition was only considered as a valid colocalization point. Figure 4
10
(c,f,i) shows that NTPDase1 immunoreactivity strongly colocalized with calnexine both in
duct cells (c), mucous acinar cells (f) and serous acinar cells (i).
11
Discussion
We studied active Ca2+ transport and ATPase activity in rat submandibular gland using an
endoplasmic reticulum enriched microsomes preparation (Hurley and Martinez 1986).
Ca2+-uptake by the microsomes was completely abolished by P-type ATPases or SERCA
inhibitors. These results give the first information about specific inhibitor actions on
submandibular Ca2+ uptake, and agree with the presence of a SERCA pump in submandibular
gland (Lee et al 1997a). However, both rate and maximal amount of accumulated Ca2+ were
lower than those measured in the well-known sarcoplasmic reticulum (SR) from rabbit
skeletal muscle (Gonzalez et al 2003). Since SR contains only the SERCA1 isoform, this
difference might result from the expression of SERCA isoforms SERCA3 and SERCA2 in the
SMG (Lee et al 1997a). It is also possible that this difference might be the result of a minor
SERCA concentration in microsomes from SMG.
High sensitivity to thapsigargin discards the possibility of a secretory pathway calcium
ATPase (SPCA) since, the pumps involved in this process are thapsigargin insensitive (Van
Baelen et al 2004; Vanoevelen et al 2005).
The mean of the work was to determine whether the ATPase activity measured in the
microsomes is only a result of a SERCA activity. In skeletal muscle, addition of calcium
ionophore A23187 to impermeable vesicles enriched in SERCA enhances the ATPase activity
since it prevents calcium accumulation and consequent ATPase inhibition (Gonzalez et al
2006). Same effect is usually found by addition of mild detergents, which produces total or
partial enzyme solubilization. However, ATPase activity in our vesicles was not modified by
ionophore addition, and it was strongly inhibited in the presence of C12E8. Furthermore,
activity was significantly inhibited by sodium azide and by suramin, two drugs that are
effective to inhibit NTPDases (Knowles and Nagy 1999; Schetinger et al 2001; Failer et al
2003; Iqbal et al 2005). Simultaneous addition of saturating concentrations of suramin and
sodium orthovanadate revealed additive effects, supporting the hypothesis of at least two
different ATP-hydrolyzing enzymes in the microsomal preparation.
When characterizing cation dependence of ATPase activity we found that, although having
more affinity for calcium, it could be also activated by magnesium or manganese when
calcium was absent as described for NTPDases (Mihaylova-Ttodorova Svletana et al 2002)
and SPCA (Van Baelen et al 2004), respectively. Km values for these cations were
significantly slower than described Km values in SERCA (Gonzalez et al 2006) or SPCA
(Van Baelen et al 2004; Vanoevelen et al 2005), but similar to those reported for NTPDases
(Mihaylova-Ttodorova Svletana et al 2002). The manganese activation was previously
12
described for different plasma membrane Mg-ATPase (Beeler et al 1985; Zhao and Dhalla
1988) but in those reports activation rates were even lower for Mn compared to Mg.
Conversely, results presented herein showed similar activation rates for both cations in
accordance with recent data obtained with a Trypanosoma Brucei NTPDase (de Souza Leite
et al 2007) .
Microsomes nucleotidase activity had a wide range of substrate specificity, presenting
significantly higher hydrolysis rate for nucleotide tri phosphates (ATP and UTP) than for
nucleotide diphosphates (ADP and UDP). Finally, ATPase activity decayed as a function of
time, as was previously described for Ca-independent Mg-ATPases (Beeler et al 1995), the
Ca/Mg ATPase of rat heart plasma membrane (Zhao and Dhalla 1988) or, more specifically,
for human NTPDase2 (Knowles and Chiang 2003).
All these results support the hypothesis that the endoplasmic reticulum microsomes from rat
submandibular gland express at least two different enzymes responsible for calcium uptake
and nucleotidase activity. The biochemical characteristics of nucleotidase activity agree with
an ecto-nucleoside triphosphate diphosphohydrolase (E-NTPDase) activity (Plesner 1995;
Heine et al 1999; Iqbal et al 2005; Kukulski et al 2005).
Eight members of E-NTPDase family have been characterized up to date in several tissues.
They hydrolyze nucleoside triphosphates and diphosphates with wide affinity range for
different substrates (Zimmermann 2001; Wu et al 2005; Kukulski et al 2005). NTPDase1
(CD39), NTPDase2 (CD39L1), NTPDase3 (CD39L3) and NTPDase8 are typical ecto-
enzymes localized in the plasma membrane (Reviewed in Grinthal and Guidotti 2006), and
they could be grouped as NTPDase1 related enzymes. NTPDase1 is the prototype member of
this group and it has been described as a detergent-sensitive dimeric active form (Wang et al
1998) that is usually located in the plasma membrane but that can be also expressed in
different cytoplasmic compartments in yeast and mammals (Zhong et al 2001; D’Alesio et al
2001; Wang et al 2005. The other four members of this enzyme family (NTPDase4-7) have a
specific intracellular localization (Wang and Guidotti 1998; Mulero et al 1999; Briederbick et
al 2000).
The presence of a NTPDase in three cellular types of the rat submandibular gland was
described using an anti-hNTPDase1 antibody (Fig. 3b-e). This antibody was able to recognize
two bands from rat submandibular gland homogenates or microsome preparation (Fig. 3a,
lanes 2-3) and was demonstrated to cross react with human platelets (Fig. 3a, lane 1).
Moreover, after deglycosylation, microsome preparation presented a third band corresponding
to deglycosylated form (Fig. 3a, lane 4). Immunohistochemistry revealed a strong
13
immunoreactivity at the plasma membrane of blood vessels as well as acinar and duct cells.
However, there was also important cytosolic anti-NTPDase1 immunoreactivity in salivary
cells. Since cytosolic NTPDase localization well correlated with the specific ER marker
calnexin (Figs. 4), it could be speculated that NTPDase described herein is one of the four
intracellular members of the NTPDase family (NTPDase4-7). The two splicing variants of
NTPDase4 (LALP70) hydrolyze a variety of nucleoside tri- and diphosphates and show
insignificant catalytic activity towards ATP or ADP as substrates (Wang and Guidotti 1998;
Biederbick et al 2000). ER- and Golgi-allocated NTPDase5 (CD39L4) and NTPDase6
(CD39L2) preferentially hydrolyze nucleoside diphosphate (Trombetta & Helenius 1999;
Mulero et al 1999; Braun et al, 2000). Finally, NTPDase7 (LALP1) has an intracellular
membranes distribution and hydrolyzes a variety of nucleoside tri- and diphosphates with a
higher preference for UTP than for ATP (Shi et al 2001). The substrate specificity of
nucleotidase activity measured in submandibular gland microsomes (Figure 2c) excludes the
hypothesis that main activity was the product of any of these intracellular isoforms. The other
members of NTPDase family are usually found in the plasma membrane. However, Zhong et
al. demonstrated that, when transfected with a CD39-like apyrase, Cos7 cells presented this
enzyme in the endoplasmic reticulum (Zhong et al 2001). It should be taking into account that
maturation process of several plasma membrane proteins occurs in the ER. These authors
showed an incomplete glycosylation pattern and a reduced activity (Zhong et al 2001).
Present results support the hypothesis that this may be the case in ER enriched fractions from
rat SMG. Indeed, (i) the NTPDase activity of SMG microsomal ER-enriched fraction was ten
fold lower than recently reported NTPDase activity from rat SMG acinar cells (Henz et al
2006) and (ii) we show the presence of a 54 kDa band corresponding to a deglycosylated form
of the enzyme in the ER, compared with the 78 kDa band (glycosylated protein) which is the
most abundant form in the SMG whole homogenate (Fig. 3a, lanes 2-4). Suramin inhibition
pattern and substrate specificity could indicate the presence of an NTPDase2 or NTPDase3
(Kukulski et al 2005; Iqbal et al 2005), however, biochemical characterization of NTPDase
activity is not enough to distinguish among NTPDases 1, 2, 3 or 8.
The question about the putative function of this NTPDase in the ER of SMG remains open
since, on one hand, it was proposed that high nucleotidase activity in the intracellular
compartments could deplete ATP and potentially inhibits other ATP-dependent components
and processes (Braakman et al 1992), but on the other hand, the presence of an active
intracellular NTPDase activity was associated with several secretory processes. It was found
in secreted vesicles observed in the lumen of pancreatic acini (Beaudoin et al 1986) as well as
14
in the pancreatic juice (Yegutkin et al 2006), the oviduct (Rosenberg et al 1977; Knowles and
Nagy 1999) and the prostatic fluid (Ronquist and Brody 1985). NTPDases could play an
important role in vesicle trafficking and protein glycosylation (Trombetta and Helenius 1999;
Beaudoin et al 1986) by modifying substrate concentrations regulating activities of enzymes
involved in these processes. This hypothesis supports NTPDase distribution in SMG, where
ductal and mucus acinar cells are directly involved in the secretory process of highly
glycosylated mucins. Another non exclusive hypothesis is that functional NTPDases from
submandibular ER might play an active role in regulating ATP concentration in duct saliva. It
was recently demonstrated that ATP is secreted by pancreatic acinar cells and that ATP
concentration is modified during its passage through the ductal system by the action of
NTPDase-like enzymes secreted to ductal lumen (Yegutkin et al 2006). Luminal ATP
concentration is crucial to control calcium signal triggered by purinergic receptor (P2R).
Indeed, it was demonstrated both in pancreatic and salivary acinar and duct cells, that
expression and sensitive of P2R were heterogenic (Lee et al 1997c, Li et al 2003, Yegutkin et
al 2006), with predominant expression of ATP/UTP sensitive receptors than ADP/UDP
sensitive receptors in duct cells.
Our results demonstrated that both, SERCA and NTPDase like enzymes are present in
submandibular gland ER. The former is responsible to calcium uptake restoring cytosolic
calcium to baseline levels before stimulation. The later could be involved in the control of
ductal ATP concentration which, in turns, would modulate purinergic signaling cascade.
15
Acknowledgements
This work was supported by the Institut National de la Santé et de la Recherche Médicale
(INSERM), the Centre National de la Recherche Scientifique (CNRS), and grants from the
Universidad de Buenos Aires (OD027) and the CONICET from Argentina (PIP02503). M.A.
Ostuni was the recipient of doctoral fellowship from the Universidad de Buenos Aires and a
postdoctoral fellow from the INSERM. The authors are grateful to Dr. B. Vidic for helpful
discussions and careful reading of the manuscript, Dr. E. Ogier-Denis for kindly providing
anti-Calnexin antibody, and Dr. M. Jandrot-Perrus for kindly providing human platelets
lysate.
16
References ALONSO GL, BAZERQUE PM, ARRIGO DM, TUMILASCI OR: Adenosine triphosphate--
dependent calcium uptake by rat submaxillary gland microsomes. J Gen Physiol 58: 340-350,
1971.
BAGINSKI ES, FOA PP, ZAK B: Microdetermination of inorganic phosphate,
phospholipids, and total phosphate in biologic materials. Clin Chem 13: 326-332, 1967.
BEAUDOIN AR, VACHEREAU A, GRONDIN G, ST JEAN P, ROSENBERG MD,
STROBEL R: Microvesicular secretion, a mode of cell secretion associated with the presence
of an ATP-diphosphohydrolase. FEBS Lett 203: 1-2, 1986.
BEELER TJ, WANG T, GABLE K, LEE S: Comparison of the rat microsomal Mg-ATPase
of various tissues. Arch Biochem Biophys 243: 644-654, 1985.
BERGERON JJ, BRENNER MB, THOMAS DY, WILLIAMS DB: Calnexin: a membrane-
bound chaperone of the endoplasmic reticulum. Trends Biochem Sci 19: 124-128, 1994.
BIEDERBICK A, KOSAN C, KUNZ J, ELSASSER HP: First apyrase splice variants have
different enzymatic properties. J Biol Chem 275: 19018-19024, 2000.
BRAAKMAN I, HELENIUS J, HELENIUS A: Role of ATP and disulphide bonds during
protein folding in the endoplasmic reticulum. Nature 356: 260-262, 1992.
BRAUN N, FENGLER S, EBELING C, SERVOS J, ZIMMERMANN H: Sequencing,
functional expression and characterization of rat NTPDase6, a nucleoside diphosphatase and
novel member of the ecto-nucleoside triphosphate diphosphohydrolase family. Biochem J
351: 639-647, 2000.
D'ALESSIO C, TROMBETTA ES, PARODI AJ: Nucleoside diphosphatase and
glycosyltransferase activities can localize to different subcellular compartments in
Schizosaccharomyces pombe. J Biol Chem 278: 22379-22387, 2003.
DE SOUZA LEITE M, THOMAZ R, FONSECA FV, PANIZZUTTI R, VERCESI AE,
MEYER-FERNANDES JR: Trypanosoma brucei brucei: biochemical characterization of
ecto-nucleoside triphosphate diphosphohydrolase activities. Exp Parasitol 115: 315-323,
2006.
DOWD FJ, PASIENIUK JA, HAND AR, CHEUNG PH, HAINES DW: Characteristics of
anion-stimulated Mg-ATPase from rat parotid gland secretory granules. Arch Oral Biol 34:
167-176, 1989.
DOWD FJ, LI LS, CAMPBELL JE, CHEUNG PH: Localization and characterization of a
parotid Ca(2+)-dependent ecto-ATPase. Crit Rev Oral Biol Med 4: 415-419, 1993.
17
DOWD FJ, MURPHY HC, LI L: Metabolism of extracellular ATP by rat parotid cells. Arch
Oral Bio; 41: 855-862, 1996
DOWD FJ, LI LS, ZENG W: Inhibition of rat parotid ecto-ATPase activity. Arch Oral Biol
44: 1055-1062, 1999.
FABIATO A, FABIATO F: Calculator programs for computing the composition of the
solutions containing multiple metals and ligands used for experiments in skinned muscle
cells. J Physiol (Paris) 75: 463-505, 1979.
FAILER BU, ASCHRAFI A, SCHMALZING G, ZIMMERMANN H: Determination of
native oligomeric state and substrate specificity of rat NTPDase1 and NTPDase2 after
heterologous expression in Xenopus oocytes. Eur J Biochem 270: 1802-1809, 2003.
GONZALEZ DA, OSTUNI MA, LACAPERE JJ, ALONSO GL: A model accounting for the
simultaneous transport of calcium and manganese in sarcoplasmic reticulum membranes. Ann
N Y Acad Sci 986: 320-322, 2003.
GONZALEZ DA, OSTUNI MA, LACAPERE JJ, ALONSO GL: Stoichiometry of ATP and
metal cofactor interaction with the sarcoplasmic reticulum Ca(2+)-ATPase: a binding model
accounting for radioisotopic and fluorescence results. Biophys Chem 124: 27-34, 2006.
GRINTHAL A, GUIDOTTI, G: CD39, NTPDase 1, is attached to the plasma membrane by
two transmembrane domains. Why? Purinergic. Signal. 2: 391-398, 2006.
HENZ SL, RIBEIRO CG, ROSA A, CHIARELLI RA, CASIALI EA, SARKIS JJ: Kinetic
characterization of ATP diphosphohydrolase and 5’-nucleotidase activities in cells cultured
from submandibular salivary glands of rats. Cell Biol Int 30: 214-220, 2006.
HURLEY TW, RYAN MP. The control of cytosolic Ca2+ concentration: studies of high
affinity Ca2+ transport in permeabilized acini of rat submandibular glands. Arch Oral Biol 33:
793-800, 1988.
HURLEY TW, MARTINEZ JR: Characterization of the kinetic and regulatory properties of
high-affinity Ca2+-ATPase activity in acinar preparations of rat submandibular salivary
glands. Arch Oral Biol 30: 587-594, 1985.
HURLEY TW, MARTINEZ JR: Characterization and localization of two forms of active
Ca2+ transport in vesicles derived from rat submandibular glands. Cell Calcium 7: 49-59,
1986.
IQBAL J, VOLLMAYER P, BRAUN N, ZIMMERMANN H, MULLER CE: A capillary
electrophoresis method for the characterization of ecto-nucleoside triphosphate
diphosphohydrolases (NTPDases) and the analysis of inhibitors by in-capillary enzymatic
microreaction. Purinergic. Signal. 1: 349-358, 2005.
18
KITTEL A, PELLETIER J, BIGONNESSE F, GUCKELBERGER O, KORDAS K, BRAUN
N, ROBSON SC, SEVIGNY J: Localization of nucleoside triphosphate diphosphohydrolase-1
(NTPDase1) and NTPDase2 in pancreas and salivary gland. J Histochem Cytochem 52: 861-
871, 2004.
KNOWLES AF, NAGY AK: Inhibition of an ecto-ATP-diphosphohydrolase by azide. Eur.
J. Biochem. 262: 349-357, 1999.
KNOWLES AF, CHIANG W-C: enzymatic and transcriptional regulation of human ecto-
ATPase/E-NTPDase 2. Arch Biochem Biophys 418: 217-227, 2003.
KUKULSKI F, LEVESQUE SA, LAVOIE EG, LECKA J, BIGONNESSE F, KNOWLES
AF, ROBSON SC, KIRLEY TL, SEVIGNY J: Comparative hydrolysis of P2 receptor
agonists by NTPDases 1, 2, 3 and 8. Purinergic. Signal. 1: 193-204, 2005.
LAEMMLI UK: Cleavage of structural proteins during the assembly of the head of
bacteriophage T4. Nature 227: 680-685, 1970.
LEE MG, XU X, ZENG W, DIAZ J, KUO TH, WUYTACK F, RACYMAEKERS L,
MUALLEM S: Polarized expression of Ca2+ pumps in pancreatic and salivary gland cells.
Role in initiation and propagation of [Ca2+]i waves. J Biol Chem 272: 15771-15776, 1997a.
LEE MG, XU X, ZENG W, DIAZ J, WOJCIKIEWICZ RJ, KUO TH, WUYTACK F,
RACYMAEKERS L, MUALLEM S: Polarized expression of Ca2+ channels in pancreatic
and salivary gland cells. Correlation with initiation and propagation of [Ca2+]i waves. J. Biol.
Chem. 272: 15765-15770, 1997b.
LEE MG, ZENG W, MUALLEM S: Characterization and localization of P2 receptors in rat
submandibular gland acinar and duct cells. J Biol Chem. 272: 32951-3295, 1997c.
LEMMENS R, KUPERS L, SEVIGNY J, BEAUDOIN AR, GRONDIN G, KITTEL A,
WAELKENS E, VANDUFFEL L: Purification, characterization, and localization of an ATP
diphosphohydrolase in porcine kidney. Am J Physiol Renal Physiol 278: F978-F988, 2000.
LI Q, LUO X, ZENG W, MUALLEM S: Cell-specific behavior of P2X7 receptors in mouse
parotid acinar and duct cells. J Biol Chem. 278: 47554-47561, 2003.
MIHAYLOVA-TODOROVA ST, TODOROV LD, WESTFALL DP: Enzyme kinetics and
pharmacological characterization of nucleotidases releasedfrom the guinea pig isolated vas
deferens during nerve stimulation: evidence for a soluble ecto-nucleoside triphosphate
diphosphohydrolase-like ATPase and a soluble ecto-5'-nucleotidase-like AMPase. J
Pharmacol Exp Ther 302: 992-1001, 2002.
19
MULERO JJ, YEUNG G, NELKEN ST, FORD JE: CD39-L4 is a secreted human apyrase,
specific for the hydrolysis of nucleoside diphosphates. J Biol Che; 274: 20064-20067, 1999.
MURPHY HC, HAND AR, DOWD FJ: Localization of an ecto-ATPase/cell-CAM 105 (C-
CAM) in the rat parotid and submandibular glands. J Histochem Cytochem 42: 561-568,
1994.
PLESNER L: Ecto-ATPases: identities and functions. Int Rev Cytol 158: 141-214,1995.
ROBSON SC, SEVIGNY J, ZIMMERMANN H: The E-NTPDase family of
ectonucleotidases: Structure function relationships and pathophysiological significance.
Purinergic. Signal. 2: 409-430, 2006.
RONQUIST G, BRODY I: The prostasome: its secretion and function in man. Biochim
Biophys Acta 822: 203-218, 1985.
ROSENBERG MD, GUSOVSKY T, CUTLER B, BERLINER AF, ANDERSON B:
Purification and characterization of an extracellular ATPase from oviductal secretions.
Biochim Biophys Acta 482: 197-212, 1977.
SCHETINGER MR, VIEIRA VL, MORSCH VM, BALZ D: ATP and ADP hydrolysis in
fish, chicken and rat synaptosomes. Comp Biochem. Physiol B Biochem. Mol. Biol. 128: 731-
741, 2001.
SHI JD, KUKAR T, WANG CY, LI QZ, CRUZ PE, DAVOODI-SEMIROMI A, YANG P,
GU Y, LIAN W, WU DH, SHE JX: Molecular cloning and characterization of a novel
mammalian endo-apyrase (LALP1). J Biol Chem 276: 17474-17478, 2001.
TEO TS, THIYAGARAJAH P, LEE MK, SELWYN MJ: The high-affinity Ca(2+)-ATPase
from rat parotid plasma membranes is an ectoenzyme: solubilization and characterization of
the Ca(2+)-ATPase activity. Biochem Med Metab Bio; 50: 358-362, 1993.
TROMBETTA ES and HELENIUS A: Glycoprotein reglucosylation and nucleotide sugar
utilization in the secretory pathway: identification of a nucleoside diphosphatase in the
endoplasmic reticulum. EMBO J 18: 3282-3292, 1999.
VALENZUELA MA, LOPEZ J, DEPIX M, MANCILLA M, KETTLUN AM, CATALAN L,
CHIONG M, GARRIDO J, TRAVERSO-CORI A: Comparative subcellular distribution of
apyrase from animal and plant sources. Characterization of microsomal apyrase. Comp
Biochem Physiol B 93: 911-919, 1989.
VAN BAELEN K, DODE L, VANOEVELEN J, CALLEWAERT G, DE SMEDT H,
MISSIAEN L, PARYS JB, RAEYMAEKERS L, WUYTACK F: The Ca2+/Mn2+ pumps in
the Golgi apparatus. Biochim Biophys Acta 1742: 103-112, 2004.
20
VANOEVELEN J, DODE L, VAN BAELEN K, FAIRCLOUGH RJ, MISSIAEN L,
RAEYMAEKERS L, WUYTACK F: The secretory pathway Ca2+/Mn2+-ATPase 2 is a
Golgi-localized pump with high affinity for Ca2+ ions. J Biol Chem 280: 22800-22808, 2005.
WANG CJ, VLAJKOVIC SM, HOUSLEY GD, BRAUN N, ZIMMERMANN H, ROBSON
SC, SEVIGNY J, SOELLER C, THORNE PR: C-terminal splicing of NTPDase2 provides
distinctive catalytic properties, cellular distribution and enzyme regulation. Biochem J 385:
729-736, 2005.
WANG TF, GUIDOTTI G: Golgi localization and functional expression of human uridine
diphosphatase. J Biol Chem 273: 11392-11399, 1998.
WANG TF, OU Y, GUIDOTTI G: The transmembrane domains of ectoapyrase (CD39) affect
its enzymatic activity and quaternary structure. J Biol Chem 273: 24814-24821, 1998.
WU JJ, CHOI LE, GUIDOTTI G: N-linked oligosaccharides affect the enzymatic activity of
CD39: diverse interactions between seven N-linked glycosylation sites. Mol Biol Cell 16:
1661-1672, 2005.
YEGUTKIN GG, SAMBURSKI SS, JALKANEN S, NOVAK I: ATP-consuming and ATP-
generating enzymes secreted by pancreas. J Biol Chem. 281: 29441-29447, 2006.
ZHAO D, DHALLA NS: Characterization of rat heart plasma membrane Ca2+/Mg2+
ATPase. Arch Biochem Biophys 263: 281-292, 1988.
ZHONG X, MALHOTRA R, WOODRUFF R, GUIDOTTI G: Mammalian plasma membrane
ecto-nucleoside triphosphate diphosphohydrolase 1, CD39, is not active intracellularly. The
N-glycosylation state of CD39 correlates with surface activity and localization. J Biol Chem
276: 41518-41525, 2001.
ZIMMERMANN H. Purinergic and Pyrimidergic Signalling I. In Abbracciio MP and
Williams M (ed). Handbook of Experimental Pharmacology. Srpinger Verlag, Heidelber
209-250, 2001.
21
Table 1. Effects of inhibitors on ATP hydrolysis by rat submandibular microsomes.
Inhibitor ATPase Activity
(% of Control)
Thapsigargin (2 µM) 89.7 ± 12.6
Sodium orthovanadate (1 mM) 89.8 ± 7.8
Ouabain (0.1 mM) 95.5 ± 10.5
Compound 48/80 (1 mg/ml) 89.3 ± 9.3
Sodium azide (10 mM) 67.4 ± 11.5*
Suramin (3 mM) 19.7 ± 2.0**
Suramin (3 mM) + sodium orthovanadate (1mM) 13.4 ± 2.9**
C12E8 (detergent/protein = 3 w/w) 28.0 ± 2.0**
Ionophore (A23187 5 µM) 98.0 ± 4.5
Control ATPase activity (100%) was 1.24 ± 0.09 µmol Pi/mg protein/min. Results are
expressed as mean ± S.E.M. (n = 5). Data were analyzed statistically by one-way analysis of
variance (ANOVA). * p< 0.05 ** p< 0.01
22
Figure Legends Figure 1: ER origin and calcium uptake by rat submandibular microsomes (a) Electron
micrograph of a thin section of the microsome preparation showed typical rough microsomes
(arrows indicated the ribosomes bound to the vesicles, bar = 200 nm). (b) Western blot for
identification of Calnexin. Immunoreactivity of microsomes from rat submandibular
endoplasmic reticulum with a rabbit anti-calnexin monoclonal antibody. (c) Calcium uptake
was measured at 37 °C, pH 7.2, 50 µM 45Ca(CaCl2), 100 mM KCl, 3 mM MgCl2, with
(triangles) or without (circles) 3 mM ATP. As indicated by arrow, 5µM calcium ionophore
A23187 was added. (d) Experiments were made with identical control condition (triangles)
but in the presence of 1 mM sodium orthovanadate (squares) or 2 µM thapsigargin (circles).
Each point is a result of at least 5 independent experiments. Results are expressed as mean ±
S.E.M.
Figure 2: ATPase activity in rat submandibular microsomes. Cation dependence, inhibitor’s
action, substrate specificity and time dependence (a) ATPase activity was plotted as function
of [Ca2+] (triangles) without added Mg, or as function of [Mg2+] (filled squares) or [Mn2+]
(open squares), without added Ca. Experiments were made at 37 ºC with 3 mM ATP. Inset:
Linewever-Burk double reciprocal plots. (b) Effect of suramin (circles, IC50 = 55 ± 10 µM)
and sodium azide (triangles, IC50 = 585 ± 160 µM) on nucleotidase activity. (c) Substrate
specificity of nucleotidase activity. Control ATPase activity (100%) was 1.24 ± 0.09 µmol/mg
protein/min. All substrates were used at 3 mM (0.1 mM Ca2+). (d) ATPase activities were
calculated as dPi/dt values from the derivative of the curve obtained for Pi as a function of
time. Each point is a result of at least 5 independent experiments. Results are expressed as
mean ± S.E.M. * P < 0.05; *** P<0.001
Figure 3: CD39 immunostaining in rat submandibular gland. Panel a: cross-reactivity of anti
hNTPDase1 monoclonal antibody was revealed by Western Blot showing positive reactivity
with human platelet lysate (lane 1), rat SMG whole homogenate (lane 2), rat SMG
microsomes (lane 3) and rat SMG microsomes after deglycosylation by PNGase (lane 4).
Panel b: Brown color of peroxydase-antiperoxydase (PAP) revealed the anti-hNTPDase1
immunoreactive cells in rat SMG counterstained with haematoxylin. Immunoreactivity is
present in blood vessels (bv), secretory ducts (du) and mucous acini (ma) but only in some
serous acini (sa). Panels c, d and e show cellular localization with a higher magnification,
23
arrow heads indicates plasma membrane strong immunoreactivity. All duct cells were
immunoreactive, with apical area strongly labeled (c). Depleted mucous cells (d) and acinar
serous cells (e) were uniformly stained. Bars: (b) = 100 µm, (c, d, and e) = 10 µm. L = lumen
Figure 4: CD39 and calnexin colocalization by confocal microscopy immunofluorescence.
Calnexin immunostaining was revealed by an anti-calnexin Ab coupled to Alexa Fluor 488
(green a, d, g) and that of CD39 by an anti-CD39 monoclonal Ab coupled to Alexa 546 (red
b, e, h). Cytoplasmic Anti calnexin immunoreactivity was found in duct cells (a), mucous
acinar cells (d) and serous acinar cells (g). Duct (b), mucous (e) and serous (h) cells were anti
CD39 immunostained. White points in figures c, f, and i indicate that both proteins colocalize
at a narrow perinuclear area. L = lumen N = nucleus, SG = secretory granules. Bar = 5 µm.
24
88 kD
200 nma b
0
10
20
30
0 10 20 30 40 50 60
Time (min)
45C
a up
take
d (n
mol
/mg)
A23187 5 µM
c
0
10
20
30
0 10 20 30 40 50
Time (min)
45C
a U
ptak
e (n
mol
/mg)
d
Figure 1
0
20
40
60
80
100
120
[Inhibitor] (log M)
ATP
ase
activ
ity,
(% o
f con
trol
)
-7 -6 -5 -4 -3 -2 -1
b-7 -6 -5 -4 -3 -2 -1-8-9
a0,0
0,2
0,4
0,6
0,8
1,0
1,2
[Cation] (log M)
ATP
ase
Act
ivity
(µm
ol/m
g pr
ot/m
in)
0
2
4
6
-0,05 0 0,05
1/[cation]
1/V
d
0,2
0,4
0,6
0,8
1
1,2
1,4
1,6
1,8
0 120 240 360 480 600
time (sec)
Vt,
(dPi
/dt)
0
20
40
60
80
100
Nuc
leot
idas
e ac
tivity
, (%
of c
ontr
ol)
*** ***
*c
ATP ADP UTP UDP
Figure 2
78 kD
54 kD
28 kD
ed
L
c
L
b
du
ma
sa
a
1 2
bv
3 4
Figure 3
N
L
a b cN
N
N
SG
L
d e f
N
N
N
g h i
Figure 4